Seismic loss estimation studies require fragility curves which are usually derived using ground motion datasets. Ground motion records can be either in the form of recorded or simulated time histories compatible with regional seismicity. The main purpose of this study is to investigate the use of alternative ground motion datasets (simulated and real) on the fragility curves. Simulated dataset is prepared considering regional seismicity parameters corresponding to Erzincan using the stochastic finite-fault technique. In addition, regionally compatible records are chosen from the NGA-West2 ground motion database to form the real dataset. The paper additionally studies the effects of hazard variability and two different fragility curve derivation approaches on the generated fragility curves. As the final step for verification purposes, damage states estimated for the fragility curves derived using alternative approaches are compared with the observed damage levels from the 1992 Erzincan (Turkey) earthquake (Mw=6.6). In order to accomplish all these steps, a set of representative masonry buildings from Erzincan region are analyzed using simplified structural models. The results reveal that regionally simulated ground motions can be used alternatively in fragility analyses and damage estimation studies.